Animals read health status from body odor and adjust whether they approach. Experiments that make healthy humans temporarily sick, usually by injecting a bacterial fragment, suggest we do something similar. Those are induced illnesses rather than real ones. So odor samples were taken from 20 donors twice, once healthy and once with a genuine , and 80 raters tried to pick out the sick one.
Just above chance, and only just. Across paired sick and healthy samples the raters identified the sick odor significantly more often than chance would give, at 56.7 percent against the 50 percent a coin would deliver. A rater’s sex and made no difference to accuracy. There was some indication that a larger change in the donor’s body temperature, though not their symptoms, improved detection.
Body odor conveys information about health status between animals and influences whether they approach or avoid one another. Experiments that induce sickness in otherwise healthy humans, usually with an injected bacterial component, suggest people can detect sensory cues of infection in others too. An induced illness is not an illness, though.
This study used real ones. samples were collected from 20 donors twice, once while healthy and once while sick with a naturally occurring acute respiratory infection. Then 80 raters, under double-blind, conditions, were asked to identify the sick body odor from each of the 20 paired samples.
Sickness detection was significantly above chance, and the magnitude of the effect was low, at 56.7 percent. Chance is 50 percent, so the signal exists and is faint. Raters’ sex and disgust sensitivity were not associated with accuracy, which removes two plausible moderators.
One thing did seem to help. There was some indication that a greater change in the donor’s body temperature between their sick and healthy states improved detection accuracy, while their reported symptoms did not. The authors conclude humans can likely use sickness odor cues to guide behaviors that reduce contagion risk, such as social avoidance. Read the number before the conclusion: 56.7 percent is a statistical result, not a usable sense.
wat. Asked which armpit sample came from someone with an infection, raters got it right 57 times in a hundred, where chance gives 50.
wat wat
A real infection is a harder and more relevant test than an injected imitation of one. The size of the effect is the finding to carry: this is a faint signal, not a diagnostic sense.
How do we know?
Tognetti and colleagues collected axillary odor samples from 20 donors while healthy and while sick with an acute respiratory infection. Then 80 raters tried to identify the sick sample from each of the 20 pairs, under double-blind forced-choice conditions.
What is still unknown
An accuracy of 56.7 percent is barely above a coin toss, so nothing here supports smelling illness in practice. Twenty donors, one infection type, and odor collected from the armpit under laboratory conditions.
Learn something?
Where to next
Can you tell where a smell is coming from with your nose alone?smell
Can you catch somebody else’s fear by smelling it?smell
Why does a chili pepper feel hot when it never warms your mouth?senses